Possibility in using computational screening of binding partners of therapeutic molecules to identify potential side effects of drugs

Drug discovery and development is fraught with challenges that makes the overall process time-consuming and expensive. Chief amongst these is the tendency of many promising drugs to fail at late stage clinical trials when the target compounds are being tested for safety and efficacy in patients. One key reason accounting for late stage failure of drug candidates is an unacceptable side effect profile. Given the cost of drug development and clinical trials, the goal in pharmaceutical industry has been the early screening of drug candidates with poor side effect profile to help focus development work on the most promising drug compounds. A potential approach for doing so might be the use of computational screening utilizing structural biology tools to probe for a drug’s side effect profile. Side effects can be described at the molecular level by the binding affinity of a molecule to different proteins and biomolecules in body fluids. Hence, the essence of the computational screening effort is in understanding (i) the potential of a drug molecule in binding other biomolecules, and (ii) the strength of the binding. But, the approach requires structural models of drug candidate and potential binding partners in the body, which may be a key handicap of the approach. In general, structural models for small molecules are more readily available compared to those of protein-based biologic drugs. More importantly, while the structures of many proteins and biomolecules are available in public databases, there remains a general lack of understanding of the structures of many biomolecules, especially those of proteins at atomic resolution. But, computational molecular docking studies remain useful in exploring protein-protein interactions between biologic drug candidates and proteins, albeit at a scale smaller than the entire proteome of human cells. Results emanating from such studies could inform and guide further drug development efforts, such as in selecting particular motifs as anchor in the scaffold of drugs given their low propensity in engaging in protein-protein interactions that could result in side effects. Lack of structural models of many proteins would result in an incomplete understanding of protein-protein or small molecule-protein interactions, which necessitate the use of experiments using cell lines for probing the safety, toxicity and efficacy of drug candidates. But, experimental approaches also face challenges such as the utility of biomarkers useful for providing readout of toxicity effect of drugs, where a protein-protein interaction resulting in an otherwise tolerable side effect may be misclassified as a toxic effect that pushes a drug candidate out of the drug development pipeline. Hence, though limited in scope, computational screening of side effect profiles of drug candidates can identify molecules that should not be taken further in drug development; thereby, reducing cost and effort in drug discovery.